AI it appears is not only capable of operating on its own after initial human instructions but it can also find weaknesses and escape its testing limits.  

An OpenAI model reportedly escaped its testing environment and hacked into another company. Should we be concerned about AI systems acting beyond human control? 

Philip Clark and listeners to Nightlife are joined by Dan Lambert, Associate Director, Science Education, ANU, with the latest from the world of science and space.

Arts, Culture and Entertainment, Community and Society, Defence and National Security, Business, Economics and Finance, Human Interest, Science and Technology, Information and Communication